Granted Patent B2

US 9,454,805 · App. 14/496,401 · Granted Sep 27, 2016

Abstract

Claims (55)

1. An apparatus for reducing noise in an image, the apparatus comprising at least one processor to implement:

a temporal noise reduction unit configured to remove temporal noise from an input frame by adjusting a temporal noise reduction intensity of the input frame based on a pixel-based motion degree and a block-based motion degree that are detected from the input frame and a reference frame, thereby to generate a first output frame; and

a spatial noise reduction unit configured to remove spatial noise from the first output frame by adjusting a spatial noise reduction intensity of the first output frame based on similarity between peripheral pixels of the first output frame and the pixel-based motion degree and the block-based motion degree, thereby to generate a final output frame,

wherein the spatial noise reduction unit comprises:

a pattern matching unit configured to measure similarity between a patch with respect to a current pixel and patches with respect to peripheral pixels within a predetermined window of the first output frame; and

a spatial noise filter configured to generate a second output frame by applying weights to the peripheral pixels according to the similarity between the patches and weighted-averaging, and generate the final output frame by weight-summing the first output frame and the second output frame by using a second motion weight according to the pixel-based motion degree and the block-based motion degree.

2. The apparatus of claim 1 , wherein the temporal noise reduction unit comprises:

a motion detection unit configured to detect the pixel-based motion degree and the block-based motion degree and determine a first motion weight according to the pixel-based motion degree and the block-based motion degree; and

a temporal noise filter configured to weight-sum the input frame and the reference frame by using the first motion weight, to generate the first output frame.

3. The apparatus of claim 2 , wherein the first motion weight has a value between 0 and 1, and increases in proportion to the pixel-based motion degree and the block-based motion degree.

4. The apparatus of claim 2 , wherein the motion detection unit comprises:

a first motion detection unit configured to calculate a first detection value indicating the pixel-based motion degree based on a ratio of a first filtering value obtained by low pass filtering the input frame with respect to a maximum value between the first filtering value and a second filtering value obtained by low pass filtering the reference frame;

a block splitting unit configured to split the input frame into a plurality of blocks;

a second motion detection unit configured to calculate a second detection value that is an average of first detection values of pixels included in each of the plurality of blocks, interpolate second detection values of a block including a current pixel and peripheral blocks, and calculate a third detection value indicating a block-based motion degree of the current pixel; and

a weight determining unit configured to calculate the first motion weight based on the first detection value of the input frame and a third detection value of a previous input frame.

5. The apparatus of claim 4 , wherein the weight determining unit comprises:

a first weight determining unit configured to determine a first weight corresponding to the first detection value; and

a second weight determining unit configured to determine a second weight corresponding to the third detection value,

wherein the first motion weight is calculated by a multiplication of the first weight and the second weight.

6. The apparatus of claim 1 , further comprising a reference frame generating unit configured to generate the reference frame by weight-summing the first output frame and the second output frame by using a third motion weight according to the pixel-based motion degree and the block-based motion degree.

7. The apparatus of claim 6 , wherein the third motion weight has a smaller value than that of the second motion weight.

8. The apparatus of claim 6 , wherein the second motion weight and the third motion weight have values between 0 and 1, and increase in proportion to the pixel-based motion degree and the block-based motion degree.

9. The apparatus of claim 1 , wherein the reference frame is a frame, previous to the input frame, from which temporal noise and spatial noise are removed.

10. An apparatus for reducing noise in an image, the apparatus comprising at least one processor to implement:

a spatial noise reduction unit configured to remove spatial noise from an input frame by adjusting a spatial noise reduction intensity of the input frame based on similarity between peripheral pixels of the input frame and a pixel-based motion degree and a block-based motion degree that are detected from the input frame, thereby to generate a first output frame, and

a temporal noise reduction unit configured to remove temporal noise from the first output frame by adjusting a temporal noise reduction intensity of the first output frame based on the pixel-based motion degree and the block-based motion degree that are detected from the input frame and a reference frame, and to generate a final output frame,

wherein the spatial noise reduction unit comprises:

a pattern matching unit configured to measure similarity between a patch with a current pixel and patches with respect to peripheral pixels within a predetermined window of the first output frame; and

a spatial noise filter configured to generate a second output frame by applying to the peripheral pixels according to the similarity between the patches and weighted-averaging, and generate the final output frame by weight-summing the first output frame and the second output frame by using a second motion weight according to the pixel-based motion degree and the block-based motion degree.

11. A method of reducing noise in an image, the method comprising:

removing temporal noise from an input frame by adjusting a temporal noise reduction intensity of the input frame based on a pixel-based motion degree and a block-based motion degree that are detected from the input frame and a reference frame, thereby generating a first output frame; and

removing spatial noise from the first output frame by adjusting a spatial noise reduction intensity of the first output frame based on similarity between peripheral pixels of the first output frame and the pixel-based motion degree and the block-based motion degree, thereby generating a final output frame,

wherein the generating of the final output frame comprises:

measuring similarity between a patch with respect to a current pixel and patches with respect to peripheral pixels within a predetermined window of the first output frame;

generating the final output frame by weight-summing the first output frame and to the similarity between the patches and weight-averaging; and

generating the final output frame by weight-summing the first output frame and the second output frame by using a second motion weight according to the pixel-based motion degree and the block-based motion degree.

12. The method of claim 11 , wherein the generating of the first output frame comprises:

detecting the pixel-based motion degree and the block-based motion degree;

determining a first motion weight according to the pixel-based motion degree and the block-based motion degree; and

generating the first output frame by weight-summing the input frame and the reference frame by using the first motion weight.

13. The method of claim 12 , wherein the first motion weight has a value between 0 and 1, and increases in proportion to the pixel-based motion degree and the block-based motion degree.

14. The method of claim 12 , wherein the determining of the first motion weight comprises:

calculating a first detection value indicating the pixel-based motion degree based on a ratio of a first filtering value obtained by low pass filtering the input frame with respect to a maximum value between the first filtering value and a second filtering value obtained by low pass filtering the reference frame;

splitting the input frame into a plurality of blocks;

calculating a second detection value that is an average of first detection values of pixels included in each of the plurality of blocks;

interpolating second detection values of a block including a current pixel and peripheral blocks and calculating a third detection value indicating a block-based motion degree of the current pixel; and

calculating the first motion weight based on the first detection value of the input frame and a third detection value of a previous input frame.

15. The method of claim 14 , wherein the calculating of the first motion weight comprises:

determining a first weight corresponding to the first detection value;

determining a second weight corresponding to the third detection value; and

calculating the first motion weight by a multiplication of the first weight and the second weight.

16. The method of claim 11 , further comprising:

generating the reference frame by weight-summing the first output frame and the second frame by using a third motion weight according to the pixel-based motion degree and the block-based motion degree.

17. The method of claim 16 , wherein the second motion weight and the third motion weight have values between 0 and 1, and increase in proportion to the pixel-based motion degree and the block-based motion degree.

18. The method of claim 11 , wherein the reference frame is a frame, previous to the input frame, from which temporal noise and spatial noise are removed.

Assignments (8)